Signal-Correlation and Spatial Impression of Stereo- and 5.1 Surround-Recordings

Dissertation (22 MB)

Based on results from a previous analysis of correlation coefficients of stereo-recordings obtained from the most common stereo-microphone techniques an empirical study was undertaken concerning a number of popular microphone techniques that are currently used for 5.1 surround. It is the aim of this research project to compare four different 5.1 microphone techniques (OCT, DECCA, KFM, AB-PC) in respect to their ability to reproduce the sound field present in the recording space as faithfully as possible in the listening room: 1) Through A/B comparison tests a number of listeners had to grade several parameters of the surround-recordings (e.g. spatial impression, localization, sound-coloration, etc.), as well as listener preference. The data were subject to analysis of variance, cluster analysis, correlation analysis and principal component analysis. Statistically significant differences were obtained mainly for the KFM-system, which was rated worst. 2) Parallel to this, a comparison of the frequency-dependent inter-aural cross-correlation (FIACC) as well as frequency dependent coherence of artificial head recordings from the original venue were compared with the artificial head recordings made in the ‘sweet spot’ of the listening room. In order to obtain more detailed information for the surround-recordings the frequency dependent cross-correlation (FCC) and frequency dependent coherence of neighboring channels were analyzed as well. Two systems stood out among the others: A) the KFM-microphone exhibited broadband out-of-phase signal components in the artificial head recordings, as well as for some of the surround-channel pairings; B) the AB-PC-system was able to replicate the correlation-function of the original artificial head recording most accurately, which is also an indicator that this microphone technique is best in recreating the original sound field of the recording venue. Theoretical considerations concerning the importance of signal de-correlation in relation to spatial impression were verified through analysis of results of the listening test as well as the acoustic measurements. The comparison of the measurements with the results of the subjective listening test coincided well, as the KFM’s out-of-phase anomalies found a counterpart in relevant ‘sound-coloration’ and lack of ‘naturalness’, as evaluated by the test listeners.

Also, both the results of the subjective listening tests, as well as the measurements go well with the findings of a number of publications, which is an indication for their plausibility.

Based on the results of this research, a ‘Binaural Quality Index for reproduced music’ (BQIrep) has been proposed, which has been derived from the ‘Binaural Quality Index’, which is well known in the field of concert hall acoustics.

As part of the thesis the performance of a new 5.1 surround microphone technique (AB-PC: ‘AB-Polycardioid Centerfill’), developed by the researcher, has been evaluated. This microphone-technique, which is based on a ‘de-correlated’ recording approach, proved to be on a par with well established systems such as OCT and DECCA. In addition, the BPT-system (‘Blumlein-Pfanzagl-Triple’) has been proposed, mainly aimed at the recording of small sound-sources in stereo or surround, which also follows a ‘de-correlated’ recording approach. The results of this research project should be of high interest to the classical segment of the recording industry, as they are able to provide a clear indication on how to achieve more naturally sounding recordings.